Fix data mismatch issue in Vertica Connector (#2433)

awslabs · Dec 26, 2024 · b5da5fb · b5da5fb
2 parents 1c686a9 + f525c4b
commit b5da5fb
Showing 1 changed file with 8 additions and 7 deletions.
diff --git a/...vertica/src/main/java/com/amazonaws/athena/connectors/vertica/VerticaMetadataHandler.java b/...vertica/src/main/java/com/amazonaws/athena/connectors/vertica/VerticaMetadataHandler.java
@@ -369,16 +369,17 @@ public GetSplitsResponse doGetSplits(BlockAllocator allocator, GetSplitsRequest
         FieldReader fieldReaderAwsRegion = request.getPartitions().getFieldReader("awsRegionSql");
         String awsRegionSql  = fieldReaderAwsRegion.readText().toString();
 
+        List<S3Object> s3ObjectsList = getlistExportedObjects(exportBucket, queryId);
+        if (s3ObjectsList.isEmpty()) {
+            // Execute queries on Vertica if S3 export bucket does not contain objects for given queryId
+            executeQueriesOnVertica(connection, sqlStatement, awsRegionSql);
+            // Retrieve the S3 objects list for given queryId
+            s3ObjectsList = getlistExportedObjects(exportBucket, queryId);
+        }
 
-        //execute the queries on Vertica
-        executeQueriesOnVertica(connection, sqlStatement, awsRegionSql);
-
-        /*
-         * For each generated S3 object, create a split and add data to the split.
-         */
         Split split;
-        List<S3Object> s3ObjectsList = getlistExportedObjects(exportBucket, queryId);
 
+        // Create a split for each s3 object
         if(!s3ObjectsList.isEmpty())
         {
             for (S3Object s3Object : s3ObjectsList)